Use of diffusion-weighted imaging and diffusion tensor imaging in assessment of myometrial invasion in patients of endometrial carcinoma and its correlation with histopathological grading (Prospective study)

Endometrial cancer (EMC) is considered one of the most common gynecological cancers worldwide. In particular, the depth of myometrial invasion and histological grade of endometrial cancers (EMCs) are strong prognostic factors. Diffusion tensor measurements as mean diffusivity (MD) and fractional anisotropy (FA) values could be useful for assessing the depth of tumor invasion and its histological grade. The study aimed to evaluate the role of diffusion-weighted imaging (DWI) and diffusion tensor imaging in the detection of myometrial invasion in cases of endometrial carcinoma and prediction of its grade in vivo. This study included 50 female patients with pathologically proved endometrial carcinoma, and their ages ranged from 38 to 67 years; the mean age was 56.15 years (± 8.229 standard deviation “SD”). There was a significant statistical difference regarding the mean values of diffusion tensor fractional anisotropy (DT-FA), diffusion tensor mean diffusivity (DT-MD) and diffusion-weighted apparent diffusion coefficient(DW-ADC) values in differentiating between intact and infiltrated myometrium with (P value ≤ 0.001). The accuracy of DT-MD, DT-FA and DWI-ADC was 98%, 90% and 86%, respectively, in the detection of myometrial invasion. There was a statistically significant difference in the mean values of DT-FA, DT-MD and DW-ADC for differentiating endometrioid adenocarcinoma grades with the overall P values (˂0.001). The accuracy of DT-FA, DT- MD and DWI-ADC for differentiating grade 3 from grade 1 or 2 endometrioid adenocarcinoma was 94.9%, 84.6% and 74.4%, respectively. For differentiating grade 1 from grade 2 or 3 endometrioid adenocarcinoma, the accuracy of DT-FA, DT-MD and DWI-ADC was 90%, 89.7% and 84.6%, respectively. Mean DT-FA, DT-MD and DW-ADC values were inversely proportional to the degree of pathological grading with r =  − 0.867, − 0.762 and − 0.706, respectively. Diffusion tensor imaging and DWI are helpful in the assessment of myometrial invasion and have a high negative correlation with histopathological grading in patients with endometrial cancer.

Background Endometrial cancer (EMC) is considered one of the most common gynecological cancers worldwide and its incidence is increasing especially in developed countries. Among endometrial cancer cases, approximately 5 percent are women younger than 40 years of age [1]. The International Federation of Gynecology and Obstetrics (FIGO) stage, lymphovascular invasion, histological subtype and grade, and presence of lymph node metastases are important prognostic indicators in the endometrial cancer (EMC). In particular, the depth of myometrial invasion and histological grade of endometrial cancers (EMCs) are strong prognostic factors. The accurate preoperative assessment of the depth of the tumor invasion and its histologic grade will therefore have a definitive impact on the selection and planning of optimal treatment in patients with endometrial cancer (EMC) [2].
Endometrial cytology and biopsy are considered the mainstays for the diagnosis of endometrial lesions, but the difficulty is encountered in patients with stenosis of the vagina or the cervix, and they may not offer an exact definite diagnosis. In addition, they are still considered to be invasive procedures [3][4][5][6].
Hysterectomy is the ordinary treatment used for endometrial cancer resulting in sterilization. Instead, conservative therapy with progestins may be used in young patients who wish to save their fertility, which can only be done if the tumor is well-differentiated grade 1 and the disease is restricted to the endometrium. Therefore, an accurate evaluation of tumor grading is essential for the selection of the appropriate treatment [1].
Conventional magnetic resonance imaging provides adequate morphological information on the uterus, and thanks to technical advances and improvements in magnetic resonance imaging (MRI) software, diffusionweighted imaging and diffusion tensor imaging (DWI and DTI) have significantly improved the diagnostic value of MRI of the pelvic region as it provides functional information and are therefore considered to be imperceptible [7].
Diffusion-weighted imaging (DWI) is a non-invasive method based on diffusion (Brownian motion) of water molecules in the extracellular space. Diffusion tensor magnetic resonance imaging (DT-MRI) is an extension of diffusion-weighted imaging (DWI), used for characterizing microstructural organization such as density and orientation of fibrous tissue in vivo. It generates parametric maps that help visualize the different aspects of tissue microstructure (mean diffusivity, tissue anisotropy, and dominant fiber orientation) [8].
Diffusion tensor imaging (DTI) is valuable in determining uterine fiber architecture. It is therefore believed that diffusion tensor imaging(DTI) quantitative measurements as mean diffusivity (MD) and fractional anisotropy (FA) values could be useful for assessing the depth of tumor invasion and its histological grade, which are the most important predictive indicators of endometrial cancers (EMCs) [1,9].
This study aims to evaluate the role of diffusionweighted imaging (DWI), apparent diffusion coefficient (ADC) value and diffusion tensor imaging (DTI) (mean diffusivity "MD" and fractional anisotropy "FA") in the detection of myometrial invasion in cases of endometrial carcinoma and the prediction of its grade in vivo (Figs. 1,2,3,4,5,6,7,8).

Image acquisition and post-processing
The examination was done using the MR machine (Philips-Ingenia, 1.5 Tesla, Netherland) with maximum amplitude 45 mT/m; gantry aperture 700 mm and 8channel phased array coils for abdomino-pelvic studies.

Post-processing
All the images were transferred to the magnetic resonance (MR) workstation (Workspace 2.6.3.5, Philips, Netherland).The signal intensity of the endometrial lesions was evaluated on high b value images, and the apparent diffusion coefficient (ADC) was obtained from the diffusion-weighted (DW) images. The signal intensity of the lesions was recorded from the diffusion tensormean diffusivity [DT-ADC (MD)] and fraction anisotropy (DT-FA) color maps which were automatically generated.

Image analysis
All the MR images, DWI and DT images and maps were revised and evaluated in consensus by 2 radiologists (HSM and MAL) with experience of 5 and 18 years in MR imaging, respectively. Both were blinded to the clinical and pathological results. The cases of disagreement were resolved by discussion and consensus.
Moreover, the inter-observer agreement percentage (%) was measured for the 2 radiologists (both observers) as regard DW-ADC, DT-MD and DT-FA in myometrial invasion and grading of endometrial cancer, to add more reliability of the results.
On DW images, the regions of interest (ROIs) were manually drawn on the tumor region avoiding necrotic or cystic parts. In cases of myometrial invasion, additional (ROIs) were placed on the intact myometrium. The diffusion tensor parameters (DT-MD and DT-FA) were calculated by manually drawing regions of interest (ROIs) with 3-5 (ROIs) were placed on the normal or diseased endometrium in the sagittal plane image and 3 (ROIs) in the axial plane image. In cases of myometrial invasion, an additional two (ROIs) were placed on the infiltrated myometrium and another two (ROIs) on the intact myometrium on a sagittal plane image.
The average values of DW-ADC, DT-FA and DT-MD were then calculated and recorded.

Statistical analysis and data interpretation
This study is a prospective methodological study. The collected data were statistically analyzed by the IBM Statistical Product and Service Solution (IBM SPSS) Statistics for Windows, Version 22.0. Armonk, NY: IBM Corp. Qualitative data were described using numbers and percentages. Quantitative data were described using mean and standard deviation (SD) for parametric data after testing normality using the Kolmogorov-Smirnov test. The significance of the obtained results was judged at the (0.05) level. One Way Analysis Of Variance (ANOVA) test was used to compare more than 2 independent groups with Post Hoc Tukey test to detect pair-wise comparison. The Kappa test was used for the inter-observer agreement percent.
The diagnostic performance of a test or the accuracy of a test to discriminate diseased cases from non-diseased cases is evaluated using receiver operating characteristic (ROC) curve analysis. Sensitivity and specificity were detected from the curve and positive predictive value (PPV), negative predictive value (NPV) and accuracy were calculated through cross-tabulation.

Results
There were 50 female patients in the study, with age range from 38 to 67 years; the mean age was 56.15 years (SD ± 8.229). The most common age group among the studied cases was between 50-59 age group (26 cases representing 52% of total cases), and the least common was 40-49 age group (3 cases representing 6% of total cases) ( Table 1). All patients had pathologically proved endometrial carcinoma with the most common pathology was endometrioid adenocarcinoma 39 patients (about 78% of total cases), and the least common was clear cell carcinoma, and endometrial stromal sarcoma (each pathology 2 cases representing 4% of total cases) ( Table 2). There was a statistical significant difference between mean values of DT-FA and DT-ADC (MD) of different malignant endometrial tumors with P value = 0.003 and < 0.001, respectively. However, the mean values of DW-ADC for different malignant endometrial lesions show no significant statistical difference (P value was P = 0.07) ( Table 2). Table 3 shows that 28 cases of endometrial lesions had positive myometrial infiltration (representing 56% of total cases), and 22 cases had negative myometrial infiltration (representing 44% of total cases). There was a significant statistical difference regarding the mean values of DT-FA, DT-MD and DW-ADC values in differentiating between intact and infiltrated myometrium with (P value ≤ 0.001) for all parameters (Table 3). Table 4 shows that: (a) Using the cutoff DT-FA value = (0.320 × 10 −3 mm 2 /s) to detect myometrial invasion, the area under curve (AUC) was 0.969. The ROC curve shows that the sensitivity, specificity and accuracy were 85.7%, 95.5% and 90.0%, respectively.  0.869. The ROC curve shows that the sensitivity, specificity and accuracy were 82.1%, 90.9% and 86.0%, respectively. P value was ˂0.001. Table 5 shows that a statistically significant difference in the mean values of DT-FA, DT-MD and DW-ADC for differentiating endometrioid adenocarcinoma grades with the overall P value was ˂ 0.001 for all parameters.
The inter-observer correlation coefficient for variability between the two observers in detecting myometrial invasions among the studied cases was 0.96, 0.89 and 1 for DW-ADC, DT-FA and DT-MD, respectively, and in differentiating Grade 3 from Grade 1 or 2 endometrioid adenocarcinoma was 0.92, 1 and 0.91 for DW-ADC, DT-FA and DT-MD, respectively, and in differentiating Grade 2 or 3 from grade 1 endometrioid adenocarcinoma was 0.94, 0.86 and 0.85 for DW-ADC, DT-FA and DT-MD, respectively (P value ≤ 0.001 for all) ( Table 8).

Discussion
Diffusion-weighted (DW) MR imaging has a higher diagnostic accuracy in assessing the depth of myometrial invasion and the overall staging of endometrial cancer than dynamic contrast-enhanced (DCE) MR imaging, and this was confirmed by Beddy et al. [10]. Yamada et al. [11,12] proved that DTI had a role in evaluation of the depth of invasion in esophageal and gastric carcinomas.
Only few articles were published about using DTI parameters in the evaluation of myometrial invasion and grading of endometrial carcinoma. The current study was designed to investigate the role of both DW-ADC and diffusion tensor parameters (mainly DT-FA and DT-MD) in the detection of myometrial infiltration and grading of endometrial carcinoma with the most common malignant pathology were endometrioid adenocarcinoma. Similar findings were detected by Toba et al. [1], Zhang et al. [13] and Yamada et al. [14], as the most common pathology was endometrioid adenocarcinoma.   Tian et al. [15] concluded that diffusion-weighted and diffusion tensor imaging could help in the differentiation of endometrioid adenocarcinoma and serous adenocarcinoma. Also, the results of the current study showed that mean DT-FA and DT-MD values of endometrioid adenocarcinoma (0.35 ± 0.13 and 0.907 ± 0.19, respectively) were more than that of papillary serous carcinoma (0.187 ± 0.01 and 0.509 ± 0.04, respectively).
As accurate diagnosis of myometrial invasion is essential in the preoperative assessment of endometrial cancer, Deng et al. [16] reported that diffusion and ADC value could aid in the prediction of deep myometrial invasion. Also, Gil et al. [17] stated that the conjunction of diffusion-weighted imaging with T2WI had better results than the dynamic contrast study and T2WI in the evaluation of myometrial invasion. Similar findings were detected in the current study, and there was a significant statistical difference considering mean DW-ADC values between intact (1.260 ± 0.17) and invaded areas of the myometrium (0.870 ± 0.16) (P value ˂0.001).  In this study, there was a significant statistical difference regarding DT-FA mean values between intact (0.450 ± 0.07) and invaded areas of myometrium (0.260 ± 0.07) (P value ≤ 0.001).
This is compatible with Zhang et al. [13], who stated that DT-FA mean values showed significant differences between cancerous (0.41) versus non-cancerous areas (0.27) within the superficial myometrium (P value ˂ 0.001). Also, they reported that mean DT-MD could distinguish cancerous (1.16) from non-cancerous areas (1.48) within the superficial myometrium (P value ˂ 0.001).
Also, Toba et al. [1] and Yamada et al. [14] concluded that DTI might be a useful tool for diagnosing myometrial invasion of uterine endometrial cancer, both ex vivo and in vivo, respectively.
Previous studies by Zhang et al. [13] and Yamada et al. [14] confirmed that the DTI parameters were superior to the ADC in the assessment of myometrial invasion of endometrial carcinoma. This is in agreement with the results of this study, as the accuracy of DT-MD (98.0%) and DT-FA (90.0%) was superior to the accuracy of DW-ADC (86.0%) in the prediction of myometrial infiltration.
For grading of endometrial carcinoma, Nakamura et al. [18], Inoue et al. [19] and Habib et al. [20] reported that DW-ADC may have a value in differentiation of endometrioid adenocarcinoma pathologic grades and the lower ADC value is associated with the higher grade of the tumor.
The results of this study are matching with the previous studies, and there was a significant statistical difference regarding DW-ADC mean values in different endometrioid adenocarcinoma grades (G1 = 1.03 ± 0.12; G2 = 0.892 ± 0.062 and G3 = 0.824 ± 0.072) (P value ≤ 0.001).
Diffusion tensor imaging proved to have a role in grading of breast, renal cancers and brain gliomas [21][22][23][24][25][26], especially DT-FA parameter which is decreased with a higher grade of the tumor and this is mostly due to packed cells and heterogeneity [22].
Yamada et al. [14] reported that the mean DT-FA and DT-MD values of endometrioid adenocarcinoma G1, G2, and G3 were statistically different with significant (P value ≤ 0.001).
The results of this study confirmed that there is an inverse correlation between the mean values of DW-ADC, DT-MD and DT-FA; and the pathological grade of endometrioid adenocarcinoma (the higher the grade, the lower the value, and vice versa) (P value ≤ 0.001). This is in accordance with Habib et al. [20] with (P value = 0.03 for DW-ADC) and in harmony with Yamada et al. [14] who stated that there was a significant inverse correlation between DT-FA and DT-MD values and histo-pathologic grades (P value < 0.001).
The analysis of the ROC curve showed that DW-ADC, DT-FA and DT-MD were useful for grading of  endometrioid adenocarcinoma with DT-FA had more accuracy than DT-MD and DW-ADC for differentiation of G3 from G1 or 2 (94.9%) and also differentiation of G1 from G 2 or 3 endometrioid adenocarcinoma (90.0%). This is in agreement with Yamada et al. [14] who stated that DT-FA had the largest area under curve for differentiation of Grade 1 from G2 or 3 and for differentiation of Grade 3 from G1 or G2 endometrioid adenocarcinoma. They added that the use of DTI parameters, especially FA values, is useful for evaluating the degree of aggressiveness of endometrial carcinoma. Lastly, a preliminary recent study done by Ghosh et al. [27] used the whole tumor histogram texture for defining the ROI to measure the DTI parameters in evaluation of myoinvasion and type of endometrial carcinoma. They assumed that this texture approach will have more reproducibility and make the results more confident than subjective drawing the region of interests (ROIs) and added that more future extensive studies are needed for better results.

Limitation
This study pays attention to endometrioid subtype of endometrial carcinoma and its histopathological grades with small numbers of other subtypes of endometrial cancer.
Also, the relatively small numbers of overall patients participated in the study.
Further studies with a large scale of patients and adding DTI and DWI to conventional pelvic MRI with more standardization of the protocols are recommended in the evaluation of endometrial cancer patients. Also, more detailed inter-observer agreement studies are recommended.

Conclusions
Diffusion tensor parameters and diffusion ADC value are non-invasive measures that can be useful in distinguishing between intact and infiltrated myometrium and help in accurate histopathological grading of endometrial carcinoma (with both DT-MD and DT-FA had more specificity and accuracy than DW-ADC). This is mandatory for planning appropriate treatment and surgery in patients with endometrial cancer. This indicates that diffusion and diffusion tensor imaging can be applied to asses' malignant pelvic pathologies.